How to use sklearn's CountVectorizerand() to g

问题:

I use sklearn.feature_extraction.text.CountVectorizer to compute n-grams. Example:

import sklearn.feature_extraction.text # FYI http://scikit-learn.org/stable/install.html
ngram_size = 4
string = ["I really like python, it's pretty awesome."]
vect = sklearn.feature_extraction.text.CountVectorizer(ngram_range=(ngram_size,ngram_size))
vect.fit(string)
print('{1}-grams: {0}'.format(vect.get_feature_names(), ngram_size))

outputs:

4-grams: [u'like python it pretty', u'python it pretty awesome', u'really like python it']

The punctuation is removed: how to include them as separate tokens?

回答1:

You should specify a word tokenizer that considers any punctuation as a separate token when creating the sklearn.feature_extraction.text.CountVectorizer instance, using the tokenizer parameter.

For example, nltk.tokenize.TreebankWordTokenizer treats most punctuation characters as separate tokens:

import sklearn.feature_extraction.text
from nltk.tokenize import TreebankWordTokenizer

ngram_size = 4
string = ["I really like python, it's pretty awesome."]
vect = sklearn.feature_extraction.text.CountVectorizer(ngram_range=(ngram_size,ngram_size), \
                                                 tokenizer=TreebankWordTokenizer().tokenize)
print('{1}-grams: {0}'.format(vect.get_feature_names(), ngram_size))

outputs:

4-grams: [u"'s pretty awesome .", u", it 's pretty", u'i really like python', 
          u"it 's pretty awesome", u'like python , it', u"python , it 's", 
          u'really like python ,']